> > I'm using Torque with Weblogic, and I have an issue where it looks like
> the connections aren't being released into Weblogic's connection pool
> correctly.
Did you set the size of  Torque's connection pool so that it contains less
connections than the WL one? I didn't do it in Resin and the Resin pool
blocked when Torque's one asked for one more connection after the last one:
after all it is normal behavior, for Torque, not releasing connections that
are placed in the pool.

I didn't have the time to look deeply inside the way Torque gets
connections, but I'm still wondering why it's such a big issue writing an
adapter that just gets connections from a DataSource. Also: can Torque's
pool be completely skipped? I had a project where Torque was used in a
system with other parts that were getting connections directly from resin's
pool, so our customer asked us to have a single configuration point for the
pool size, so we had to force Torque to get connections from resin's pool.

 I've seen that currently there's a non-configurable part of the system
where adapters are associated to the driver class (actually, the standard
configuration file seemed to allow definitions of new drivers, but that part
of the file is not taken into account by the configurator class): I wrote a
small hack of the Oracle adapter and modified a bit the configuration class
in order to have Torque get connections from Resin's pool (using the
DriverManager and jdbc:caucho:poolName), and to be able to associate my
custom adapter to the resin's driver: if the list is interested I can submit
the code (well, the custom adapter has only a method that has been changed
from the standard one).


--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to